Compositional Boosting for Computing Hierarchical Image Structures

45 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Examining CNN representations with respect to Dataset Bias (2017, background): “(Wu, Xia, and Zhu 2007; Yang, Wu, and Zhu 2009; Wu and Zhu 2011) tried to isolate and diagnose information from local, bottom-up, or top-down inference processes.”
Evaluating information contributions of bottom-up and top-down processes (2009, methodology): “…such as AdaBoost [14], (ii) pure top-down computing, such as template matching [18], and (iii) one pass of bottom-up followed by a top-down phase in a separate manner, such as DDMCMC [11], compositional boosting algorithm [15], compositional detection method [19] and recent active basis model [16].”
RAVEN: A Dataset for Relational and Analogical Visual REasoNing (2019, methodology): “Speciﬁcally, we use the A-SIG as the representation of RPM; each RPM is a parse tree that instantiates from the A-SIG.”
Sketch Tokens: A Learned Mid-level Representation for Contour and Object Detection (2013, methodology): “We propose a novel approach to deﬁn-ing token classes using supervised mid-level information, unlike previous approaches that use hand-deﬁned classes [39], high-level supervision [16], or unsupervised information [15].”